The 3D semantic instance prediction task involves detecting and segmenting the object in an 3D scan mesh.

Evaluation and metrics

Similarly to the ScanNet benchmark in ScanNet200 our evaluation ranks all methods according to the average precision for each class. We report the mean average precision AP at overlap 0.25 (AP 25%), overlap 0.5 (AP 50%), and over overlaps in the range [0.5:0.95:0.05] (AP) for all 200 categories. Note that multiple predictions of the same ground truth instance are penalized as false positives.



This table lists the benchmark results for the ScanNet200 3D semantic instance scenario.




Method Infoavg ap 50%head ap 50%common ap 50%tail ap 50%alarm clockarmchairbackpackbagballbarbasketbathroom cabinetbathroom counterbathroom stallbathroom stall doorbathroom vanitybathtubbedbenchbicyclebinblackboardblanketblindsboardbookbookshelfbottlebowlboxbroombucketbulletin boardcabinetcalendarcandlecartcase of water bottlescd caseceilingceiling lightchairclockclosetcloset doorcloset rodcloset wallclothesclothes dryercoat rackcoffee kettlecoffee makercoffee tablecolumncomputer towercontainercopiercouchcountercratecupcurtaincushiondecorationdeskdining tabledish rackdishwasherdividerdoordoorframedresserdumbbelldustpanend tablefanfile cabinetfire alarmfire extinguisherfireplacefolded chairfurnitureguitarguitar casehair dryerhandicap barhatheadphonesironing boardjacketkeyboardkeyboard pianokitchen cabinetkitchen counterladderlamplaptoplaundry basketlaundry detergentlaundry hamperledgelightlight switchluggagemachinemailboxmatmattressmicrowavemini fridgemirrormonitormousemusic standnightstandobjectoffice chairottomanovenpaperpaper bagpaper cutterpaper towel dispenserpaper towel rollpersonpianopicturepillarpillowpipeplantplateplungerposterpotted plantpower outletpower stripprinterprojectorprojector screenpurserackradiatorrailrange hoodrecycling binrefrigeratorscaleseatshelfshoeshowershower curtainshower curtain rodshower doorshower floorshower headshower wallsignsinksoap dishsoap dispensersofa chairspeakerstair railstairsstandstoolstorage binstorage containerstorage organizerstovestructurestuffed animalsuitcasetabletelephonetissue boxtoastertoaster oventoilettoilet papertoilet paper dispensertoilet paper holdertoilet seat cover dispensertoweltrash bintrash cantraytubetvtv standvacuum cleanerventwardrobewashing machinewater bottlewater coolerwater pitcherwhiteboardwindowwindowsill
sort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysorted bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort by
Mask3D Scannet2000.314 10.480 10.264 10.171 10.500 20.024 10.500 10.000 10.000 10.000 10.362 10.265 30.366 11.000 10.853 10.865 10.500 10.004 10.250 20.188 10.000 10.178 10.320 10.443 30.345 10.000 10.069 30.000 10.000 10.143 10.348 11.000 10.000 10.000 10.000 10.758 10.121 10.000 40.000 10.167 10.167 10.162 10.333 10.000 10.000 10.192 40.758 20.250 20.833 11.000 10.845 20.000 10.109 10.650 10.143 10.000 10.692 10.125 10.533 20.200 20.728 10.377 10.689 10.500 30.126 10.143 20.741 10.000 10.017 20.400 10.000 10.063 10.000 10.119 30.744 10.500 10.522 10.328 10.397 10.611 10.543 10.000 10.000 10.342 10.000 10.147 10.036 10.167 10.042 30.400 10.804 10.250 10.496 10.887 10.429 10.000 10.515 40.049 10.039 10.083 10.000 10.000 11.000 10.661 10.028 10.052 30.500 10.490 40.566 10.000 10.944 10.083 10.519 10.000 10.000 10.000 10.097 40.000 11.000 10.000 10.679 20.667 20.000 30.664 40.000 10.000 10.273 10.262 10.250 11.000 10.500 20.938 10.050 10.538 10.000 10.707 10.784 10.227 20.111 10.150 11.000 10.000 10.500 10.000 10.000 10.000 10.649 10.000 10.000 10.363 10.497 20.342 10.333 10.000 10.817 40.359 20.677 10.449 10.629 10.863 10.000 10.000 10.645 20.667 30.050 40.000 10.000 10.528 20.000 10.333 20.126 20.040 40.540 10.468 1
Minkowski 34D Inst.permissive0.203 40.369 30.134 40.078 40.479 30.003 30.500 10.000 10.000 10.000 10.100 20.371 10.300 20.667 30.746 20.400 20.000 20.000 20.000 30.031 20.000 10.074 30.165 30.413 40.000 30.000 10.070 20.000 10.000 10.000 20.221 40.000 20.000 10.000 10.000 10.372 40.070 20.706 20.000 10.000 20.000 40.123 30.033 40.000 10.000 10.422 30.732 30.000 30.778 41.000 10.845 30.000 10.090 30.636 20.000 20.000 10.158 30.000 20.250 40.050 40.693 20.123 30.051 40.385 40.009 30.118 40.406 40.000 10.000 30.200 20.000 10.000 20.000 10.133 20.307 40.500 10.251 30.000 30.281 20.402 40.317 20.000 10.000 10.000 20.000 10.060 30.000 20.000 20.396 10.200 20.669 20.021 30.218 40.720 40.000 20.000 10.696 10.025 30.000 20.000 20.000 10.000 10.125 40.596 20.000 20.191 10.500 10.595 10.369 30.000 10.500 30.000 20.143 40.000 10.000 10.000 10.226 30.000 10.000 20.000 10.701 10.511 30.000 30.851 20.000 10.000 10.150 30.052 40.100 30.981 20.500 20.286 20.000 20.000 40.000 10.545 30.522 40.250 10.000 20.000 40.522 40.000 10.500 10.000 10.000 10.000 10.282 40.000 10.000 10.178 40.382 30.018 40.056 30.000 10.997 20.107 40.677 10.313 30.000 30.726 40.000 10.000 10.583 40.903 20.200 30.000 10.000 10.333 30.000 10.442 10.083 30.109 30.387 30.000 4
C. Choy, J. Gwak, S. Savarese: 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks. CVPR 2019
CSC-Pretrain Inst.permissive0.209 30.361 40.157 30.085 30.506 10.007 20.500 10.000 10.000 10.000 10.000 40.093 40.221 30.667 30.524 40.400 20.000 20.000 20.000 30.004 30.000 10.000 40.109 40.589 20.000 30.000 10.059 40.000 10.000 10.000 20.322 20.000 20.000 10.000 10.000 10.405 20.055 30.700 30.000 10.000 20.028 30.091 40.083 20.000 10.000 10.667 10.768 10.000 30.807 31.000 10.776 40.000 10.000 40.340 40.000 20.000 10.103 40.000 20.750 10.200 20.634 40.053 40.246 20.677 20.006 40.198 10.432 30.000 10.000 30.050 30.000 10.000 20.000 10.111 40.356 30.500 10.188 40.000 30.220 30.448 20.050 40.000 10.000 10.000 20.000 10.032 40.000 20.000 20.396 10.000 30.573 40.000 40.228 30.747 30.000 20.000 10.573 30.021 40.000 20.000 20.000 10.000 10.500 30.573 30.000 20.000 40.125 40.592 20.364 40.000 10.450 40.000 20.364 20.000 10.000 10.000 10.340 20.000 10.000 20.000 10.610 30.833 10.221 10.702 30.000 10.000 10.135 40.094 30.125 20.571 30.500 20.143 40.000 20.125 20.000 10.618 20.667 30.115 40.000 20.125 21.000 10.000 10.500 10.000 10.000 10.000 10.502 30.000 10.000 10.312 30.248 40.050 30.000 40.000 10.997 20.420 10.500 40.149 40.451 20.748 20.000 10.000 10.636 30.667 30.600 10.000 10.000 10.278 40.000 10.333 20.000 40.294 10.381 40.110 2
Ji Hou, Benjamin Graham, Matthias Nie├čner, Saining Xie: Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts. CVPR 2021
LGround Inst.permissive0.246 20.413 20.170 20.130 20.455 40.003 40.500 10.000 10.000 10.000 10.017 30.333 20.111 41.000 10.681 30.400 20.000 20.000 21.000 10.003 40.000 10.167 20.190 20.637 10.067 20.000 10.081 10.000 10.000 10.000 20.264 30.000 20.000 10.000 10.000 10.387 30.031 40.754 10.000 10.000 20.151 20.135 20.056 30.000 10.000 10.582 20.589 40.500 10.815 21.000 10.903 10.000 10.097 20.588 30.000 20.000 10.234 20.000 20.500 30.400 10.682 30.156 20.159 30.750 10.046 20.125 30.660 20.000 10.200 10.000 40.000 10.000 20.000 10.164 10.402 20.500 10.373 20.025 20.143 40.426 30.317 20.000 10.000 10.000 20.000 10.063 20.000 20.000 20.000 40.000 30.575 30.250 10.241 20.772 20.000 20.000 10.653 20.034 20.000 20.000 20.000 10.000 11.000 10.561 40.000 20.100 20.500 10.541 30.452 20.000 10.581 20.000 20.364 20.000 10.000 10.000 10.571 10.000 10.000 20.000 10.568 40.511 30.167 20.857 10.000 10.000 10.164 20.112 20.000 40.530 41.000 10.286 20.000 20.125 20.000 10.464 40.706 20.208 30.000 20.125 20.744 30.000 10.500 10.000 10.000 10.000 10.511 20.000 10.000 10.344 20.541 10.068 20.333 10.000 11.000 10.196 30.533 30.318 20.000 30.748 30.000 10.000 10.690 11.000 10.400 20.000 10.000 10.667 10.000 10.333 20.333 10.270 20.399 20.083 3
David Rozenberszki, Or Litany, Angela Dai: Language-Grounded Indoor 3D Semantic Segmentation in the Wild.